The Evidence on Teaching

Deliberate Practice: Is That All It Takes to Become an Expert?

Hambrick, D. Z., Oswald, F. L., Altmann, E. M., Meinz, E. J., Gobet, F., & Campitelli, G. · 2014

grade Creviewindependentreplicatednumbers spot-checked
Population
Chess players (intermediate to grandmaster, international samples) and musicians (beginner to professional)
Design
Reanalysis of all locatable chess (6 studies, N=1,083) and music (8 studies) datasets having both cumulative deliberate practice and an objective performance measure. Correlations DISATTENUATED for measurement error (assumed reliability .80 practice, .91 chess rating), so these are generous upper bounds. Also reports within-skill-group distributions of accumulated hours, where the 10,000-hour threshold breaks.
Key findings
Chess and music case evidence against a practice threshold. Masters' accumulated practice ranged 832-24,284 hours (M=10,530, SD=7,414); 31.3% of masters (5 of only 16) had less practice than the mean of the tier below; and 12.9% (n=4) of intermediate players had MORE practice than the master-group mean (10,530 h) while remaining intermediate. Polgar comparison: 2,735 h vs 59,904 h. Small denominators — read as existence proofs against a threshold, not as rate estimates.
Genetic confound
Not modelled, but the thesis is that the residual is heritable capacity: chess players' mean performance IQ 129 (top third 131 vs bottom third 124); working memory predicts sight-reading above DP with NO DP x WMC interaction (no compensation at high practice).
Replication notes
Its chess/music estimates sit between Macnamara's (21-26%) and Platz's (37% corrected); the spread reflects disattenuation and inclusion strictness, not data disagreement. The within-group hours spread has not been contested.
DOI / URL
10.1016/j.intell.2013.04.001

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Chess performance (Elo rating)avg disattenuated r; % varianceavg r-hat=.57 (N-weighted .49) → 34.0% of RELIABLE variance; 66% unexplained. Study range r=.33 to .69standardizedattained expertise; cross-sectional; field (Elo ratings)nonenot-applicabledomain-skill
Music performanceavg disattenuated r; % varianceavg r-hat=.52 → 29.9% of reliable variance; 70.1% unexplainedmixedattained expertise; cross-sectional; mostly field, some lab sight-readingnonenot-applicabledomain-skill
Within-skill-group spread of hours (Gobet & Campitelli 2007, n=90)range and overlap of accumulated hoursMasters M=10,530 h (SD 7,414) but range 832–24,284 h. 31.3% of masters had LESS practice than the mean of the expert group one tier down; 12.5% less than the intermediate mean. One player reached master in <2 yr, another took 26 yrstandardizedattained expertise; fieldnonenot-applicabledomain-skill
People with >10,000 h who never became expertcount4 participants in the same chess sample exceeded 10,000 h of deliberate practice and remained INTERMEDIATE-level playersstandardizedattained expertise; fieldnonenot-applicabledomain-skill
Polgar sisters (practice vs attainment)peak Elo vs estimated hoursOne sister peaked at 2735 on ~59,904 h; another at 2577 (>1 SD lower) on ~79,248 h — more practice, worse outcome. Both far exceeded top-10 grandmasters' mean of 14,020 h (SD 7,374)standardizedattained expertise; fieldnonenot-applicabledomain-skill

Cited by